To improve local search ability of artificial bee colony（ABC） algorithm and speed up its convergence rate

the shaft of Rosenbrock method is introduced into ABC so that a Rosenbrock shaft artificial bee colony（RSABC） algorithm is proposed.The modified algorithm would realize one shaft operation in the neighborhood of current optimum at regular interval of iterations for leading itself to find the fastest direction of the function value decline.Besides

in order to gain a high quality original population

the new algorithm uses opposition strategy to regulate its original population which is randomly generated.By character test of several standard test functions

the fast convergence and stability of this algorithm is validated

it shows that the improvement is feasible and effective.